A Persistent High-Energy Flux from the Heart of the Milky Way : Integral's view of the Galactic Center

HAL Id: hal-00007740

https://hal.archives-ouvertes.fr/hal-00007740

Submitted on 4 Aug 2005

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

A Persistent High-Energy Flux from the Heart of the Milky Way : Integral’s view of the Galactic Center

G. Bélanger, A. Goldwurm, M. Renaud, R. Terrier, Fulvio Melia, Niels Lund, J. Paul, Gerry Skinner, Farah Yusef-Zadeh

To cite this version:

G. Bélanger, A. Goldwurm, M. Renaud, R. Terrier, Fulvio Melia, et al.. A Persistent High-Energy

Flux from the Heart of the Milky Way : Integral’s view of the Galactic Center. The Astrophysical

Journal, American Astronomical Society, 2006, 636, pp.275-289. �hal-00007740�

ccsd-00007740, version 1 - 4 Aug 2005

Preprint typeset using L^ATEX style emulateapj v. 2/19/04

A PERSISTENT HIGH-ENERGY FLUX FROM THE HEART OF THE MILKY WAY:

INTEGRAL’S VIEW OF THE GALACTIC CENTER

G. B´ 

, A. G 

, M. R 

, R. T 

, F. M 

, N. L 

, J. P 

, G. S 

, F. Y  -Z 

(Received 2005 May 31; Accepted 2005 July 21)

2005 May 28

ABSTRACT

Highly sensitive imaging observations of the Galactic center (  ) at high energies with an angular resolution of order 10 arcminutes, is a very recent development in the field of high-energy astrophysics. The Ibis/Isgri imager on the Integral observatory detected for the first time a hard X-ray source, IGR J17456–2901, located within 1 arcminute of Sagittarius A

(Sgr A

) over the energy range 20–100 keV. Here we present the results of a detailed analysis of approximately 7 × 10

s of observations of the  obtained since the launch of Integral in October 2002.

Two years and an effective exposure of 4.7 × 10

s have allowed us to obtain more stringent positional constraints on this high-energy source and to construct its spectrum in the range 20–400 keV. Furthermore, by combining the Isgri spectrum together with the total X-ray spectrum corresponding to the same physical region around Sgr A

from XMM-Newton data, and collected during part of the gamma-ray observations, we constructed and present the first accurate wide band high-energy spectrum for the central arcminutes of the Galaxy. Our complete and updated analysis of the emission properties of the Integral source shows that it is faint but persistent with no variability above 3 σ, contrary to what was alluded to in our first paper. This result, together with the spectral characteris- tics of the soft and hard X-ray emission from this region, suggests that the source is most likely not point-like but, rather, that it is a compact, yet diffuse, non-thermal emission region. The centroid of IGR J17456–2901 is estimated to be R.A. = 17

45

42

.5, decl. = − 28

59

28

(J2000), offset by 1

from the radio position of Sgr A

and with a positional uncertainty of 1

. Its 20–400 keV luminosity at 8 kpc is L = (5.37 ± 0.21) × 10

erg s

. A 3 σ upper limit on the flux at the electron-positron annihilation energy of 511 keV from the direction of Sgr A

is set at 1.9 × 10

ph cm

s

. Very recently, the Hess collaboration presented the detection of a source of ∼ TeV γ-rays also located within an arcminute of Sgr A

. We present arguments in favor of an interpretation according to which the photons detected by Integral and Hess arise from the same compact region of diffuse emission near the central black hole and that the supernova remnant Sgr A East could play an important role as a contributor of very high-energy γ-rays to the overall spectrum from this region. There is also evidence for hard emission from a region located between the central black hole and the radio Arc near l ∼ 0.1

along the Galactic plane and known to contain giant molecular clouds.

Subject headings: black hole physics — Galaxy: center — Galaxy: nucleus — X-rays: observations — stars:

neutron — X-rays: binaries

1. INTRODUCTION

The year 2004 marked the 30

anniversary of the discovery of the compact radio source Sgr A

(Balick & Brown 1974), which is now firmly believed by many to be the manifesta- tion of a supermassive black hole that sits at the very heart of the Milky Way and around which everything in the Galaxy turns. That year also marked the first detection of γ-rays from a compact region of size ∼ 10 arcminutes around Sgr A

with the Integral observatory in the energy range from 20 to 100 keV (B´elanger et al. 2004) and with the Hess Cerenkov telescope array between 165 GeV and 10 TeV (Aharonian et al. 2004).

After three decades of observations, we finally detected a source of very high-energy radiation that appears to be point- like and coincident with the Galactic nucleus (  ). However, the exact nature of the highly energetic emission from this com-

*Based on observations with INTEGRAL, an ESA project with instruments and science data center funded by ESA member states (especially the PI coun- tries: Denmark, France, Germany, Italy, Switzerland, Spain), Czech Republic and Poland, and with the participation of Russia and the USA.

1 Service d’Astrophysique, DAPNIA/DSM/CEA, 91191 Gif-sur-Yvette, France; be- langer@cea.fr

2Unit´e mixte de recherche Astroparticule et Cosmologie, 11 place Berthelot, 75005 Paris, France

3Physics Dept. and Stewart Observatory, Univerity of Arizona, Tucson, AZ 85721, USA; melia@physics.arizona.edu

4 Danish National Space Center, Juliane Maries vej 30, Copenhagen, Denmark;

nl@spacecenter.dk

5CESR, Toulouse Cedex 4, France; skinner@.cesr.fr

6Department of Physics and Astronomy, Northwestern University, Evanston, IL 60208

pact region is unknown. Our aim here is to present observa- tional evidence that will lead to a deeper understanding of the emission process, and help to unfold the mystery of the γ-rays arriving from the heart of the Milky Way.

The Galactic nuclear region is very dense and complex — so dense that certainly more than one source could contribute to the high-energy flux detected by present-day γ-ray instru- ments with typical angular resolutions of 10–15 arcminutes.

Located at 8 kpc from the Sun, the Galactic center (  _{) har-}

bours a supermassive black hole whose presence and mass of about 3.6 × 10

M

were deduced primarily from near- infrared (  ) observations and measurements of the velocity and proper motion of the stars contained in the central cluster (Sch ¨odel et al. 2003; Ghez et al. 2005; Eisenhauer et al. 2005).

A black hole (  ) of this mass has a Schwarzschild radius (R

) of about 1.2 × 10

cm and is expected to accrete the mat- ter from its nearby environment producing a detectable emis- sion in a broad range of frequencies (Melia & Falcke 2001).

The bright, compact, non-thermal radio source Sgr A

, located at less than 0

.01 from the dynamical center of the central star cluster, is most likely the manifestation of such accretion processes. Undetectable in the visible and  bands due to the large Galactic absorption, and only recently detected in

 both in quiescent, and flaring states (Genzel et al. 2003;

Ghez et al. 2004), this source is surprisingly weak in X-rays,

where it appears slightly extended with a luminosity of only

L

≈ 2 × 10

erg s

(Baganoff et al. 2003).

The dense central region of the Milky Way where the 2–

10 keV flux is heavily dominated by diffuse radiation, in- cludes the emission from a hot (kT ∼ 6–8 keV) probably un- bound thermal plasma component, the supernova remnant (  ) Sgr A East, several knots and filaments, and thou- sands of point sources (Maeda et al. 2002; Park et al. 2004;

Muno et al. 2004a; Muno et al. 2004b). The faint X-ray coun- terpart of Sgr A

would likely have gone unnoticed were it not positioned almost exactly at the centre of the Galaxy.

Following a monitoring of this faint X-ray source by the Chandra observatory, it was discovered that Sgr A

is the site of sometimes powerful X-ray flares (Baganoff et al. 2001).

This flaring activity was also detected and studied with XMM-Newton (Goldwurm et al. 2003a; Porquet et al. 2003a;

B´elanger et al. 2005) and the peak luminosity during these flares was seen to rise above quiescence by factors up to 180 and then decay in a few hours or less. The majority of flares have spectra that are significantly harder (power law photon in- dex Γ ∼ 1.5) than the quiescent spectrum (Γ ∼ 2.7), but the most powerful of them was quite soft (Γ ∼ 2.5). Note that dust scat- tering can have some effect on the observed fluxes and spec- tral indeces of both the quiescent and flaring states of Sgr A

(Tan & Draine 2004). Variations on time scales as short as 200 s have been detected during flares indicating an emitting region with a size of the order of 10R

. Recent Chandra and XMM-Newton campaigns have allowed us to estimate the av- erage X-ray flare rate to about 1 per day bearing in mind that the flaring events appear to be clustered (B´elanger et al. 2005).

At slightly longer wavelengths,  observations with the VLT Naco imager (Genzel et al. 2003) and Keck telescope (Ghez et al. 2004) have shown that Sgr A

is also the source of frequent  flares with durations comensurable with those seen in X-rays. Some  flares even appear to be simultaneous with X-ray flares (Eckart et al. 2004). Both the  _{and X-ray}

flaring events strongly suggest the presence of an important population of non-thermal relativistic electrons in the vicin- ity of the  horizon (Markoff et al. 2001; Liu & Melia 2002;

Yuan et al. 2002; Yuan et al. 2003; Liu et al. 2004) and there- fore their detection has raised great interest in the possibility of observing hard X-rays from the Galactic nucleus.

Sgr A

’s bolometric luminosity from radio to X-rays (in- cluding the flares) barely amounts to a few 10

erg s

, while the Eddington luminosity for a  of its mass reaches L

≈ 4 × 10

erg s

. Even the expected accretion luminosity, based on an estimated stellar wind mass rate at the accretion radius, is of the order of 10

erg s

, i.e. about 6 to 7 or- ders of magnitude higher than Sgr A

’s total observed emitted power (see Cuadra et al. 2005 for a recent update on the issue).

In the 1990’s it was thought that the bulk of the power could be found at higher energies; in hard X-rays as is the case for 

binaries in the hard state, or even at the electron-positron anni- hilation energy of 511 keV (see e.g. Genzel & Townes 1987).

Other than the detections in the 10–20 keV range from the di- rection of the  _{based on}  /Spacelab 2 (Skinner et al. 1987) and Art-P/Granat (Pavlinsky et al. 1994) data, no detection at energies above 20 keV emanating from Sgr A complex was reported. A long monitoring of the region by the Sigma telescope on the Granat satellite yielded upper limits of the order of several 10

erg s

to the 35–150 keV emission from Sgr A

(Goldwurm et al. 1994; Goldoni et al. 1999), and 2.3 × 10

ph cm

s

to the flux at 511 keV from a point source at the  (Malet et al. 1995).

In the 100 MeV to 10 GeV energy range, an uniden- tified γ-ray source of the Egret/  catalog 3EG J1746–

2851, was found to be somewhat compatible with the 

(Mayer-Hasselwander et al. 1998). In the third Egret catalog (Hartman et al. 1999), this source is located 0.17

from Sgr A

and the reported error radius is 0.13

at 90% confidence level.

Taking this at face value would marginally exclude Sgr A

. However, given the 1

angular resolution of the instrument and hence its inability to exclude the contribution from other sources contained in a region of this size, 3EG J1746–2851 is still considered as possibly coincident with the  _.

Finally, Integral observations performed during the first half of 2003 with the Ibis telescope revealed for the first time the presence of a significant excess in the energy range 20–100 keV coming from the inner region of the Galaxy (B´elanger et al. 2004). The position of this excess was found to lie 1

from Sgr A

with a 4

error radius, and its luminos- ity was estimated to be L

≈ 3 × 10

erg s

. An indication of variability on timescales comparable to those of flares in Sgr A

was reported but a subsequent analysis that included data collected over the second half of 2003 with improved analysis procedures found the source to be stable (Goldwurm et al. 2004). In 2003 June, Chandra detected two new transient sources in the close vicinity of Sgr A

. Given the 12

resolution of the Ibis/Integral telescope, the possibility that the 20–100 keV excess was linked to the emission from these objects, was considered and discussed in B´elanger et al. (2004).

Meanwhile the Hess collaboration (Aharonian et al. 2004) announced the detection of a bright source of TeV photons co- incident with Sgr A

to within 1

. The source appears to be point-like, stable, with a power-law spectrum of index of 2.2 and a luminosity of L

≈ 10

erg s

.

This wealth of high-energy data all point to the presence of one, or several, high energy non-thermal emission components likely produced by accelerated particles in the environment of the  . Both leptonic and hadronic origins for the accelerated particles giving rise to the γ-rays have been considered either in the inner (Markoff et al. 1997; Aharonian & Neronov 2005) or outer (Atoyan & Dermer 2004) region of Sgr A

, while Melia et al. (1998), Fatuzzo & Melia (2003) and recently Crocker et al. (2005), discussed the possibility that the site of particle ac- celeration could be the unusual supernova remnant Sgr A East.

Sgr A East is a mixed morphology  whose center is at less than 1

from Sgr A

and whose radio shell spans a few arcminutes. Given its position and dimensions it would appear like a stable point-like source for γ-ray observato- ries. This remnant is characterised by a non-thermal ra- dio shell at the center of which lies an apparently ejecta- dominated X-ray emiting region whose spectrum indicates that the plasma is rich in heavy elements especially towards the core where Fe abundances reach 4–5 times solar (Maeda et al. 2002;

Sakano et al. 2004). For this reason, it has been classified as a metal-rich mixed morphology  . The fact that Sgr A East is the smallest of the known  s of this type and that its radio shell appears to be quite symmetrical, although slightly elon- gated along the Galactic plane, suggests that the ejecta from the explosion have expanded in a very dense but more or less homogenous environment (Maeda et al. 2002). According to a more recent analysis of the X-ray features of Sgr A East by Sakano et al. (2004) based on XMM-Newton observations, the derived total energy, mass and abundance pattern are consistent with a single supernova event of Type Ia or Type II involving a relatively low-mass progenitor star. Furthermore, according to these authors the morphology and spectral characteristics do not show evidence of a clear connection between the  _and

past activity in Sgr A

.

Sgr A East may very well have arisen from a single explosion

akin to most other supernova events in terms of its energetics

as is suggested by Sakano et al. (2004). The most recent simu- lations for its genesis and evolution (Fryer et al. 2005) indicate that the progenitor was likely a star of mass ∼ 15 M

that ex- ploded a mere ∼ 1750 years ago. Reaching the M-0.02-0.07 cloud some 300–400 years ago, the expanding shock collided with the dense molecular gas, producing a bright flash of 2–

200 keV emission lasting several hundred years, whose X-ray echo we may be viewing today in the form of Fe fluorescent emission from Sgr B2 (and other nearby clouds). However, Sgr A East does distinguish itself from other Galactic rem- nants in three important ways. First, it is located very near the

 , within 50

of Sgr A

, and is therefore subject to interac- tions and forces uncommon in the rest of the Galaxy. Second, its non-thermal shell emission caused by synchrotron radiation from relativistic electrons has an unusually high surface bright- ness compared to other Galactic  (Green 2004). Third, OH (1720 MHz) maser emission detected in several locations around the  , and particularly at the boundary of Sgr A East and the M-0.02-0.07 molecular cloud, indicates the presence of strong shocks where rapid acceleration of electrons (and pro- tons) is taking place, in a medium threaded by very strong magnetic fields of order 2–4 mG (Yusef-Zadeh et al. 1996). As pointed out by Yusef-Zadeh, Melia & Wardle (2000), the pres- ence of relativistic electrons and strong magnetic fields within Sgr A East makes it a unique and potentially powerful Galactic accelerator.

Other possible sources of non-thermal γ-rays, some of which were initially proposed to explain certain features of the X- ray emission like the hot component and the bright 6.4 keV line of neutral Fe, or the non-thermal radio characteristics of the region, include the radio Arc, several non-thermal fila- ments and regions marked by cosmic ray electron interactions (Yusef-Zadeh et al. 2002), supernova ejecta (Bykov 2002), and scattering of highly energetic radiations from molecular clouds.

In particular Revnivtsev et al. (2004) proposed a model in which the Integral source IGR J17475–2822, coincident with the dense molecur cloud Sgr B2, is due to Compton reflec- tion (i.e. the scattering of high-energy photons by cold elec- trons in the outer layers of the cloud) of very energetic emis- sion from a very powerful flare in the Sgr A

system about 300 years ago. In any case, none of the current models can inte- grate the three high-energy sources (IGR J17456–2901, 3EG J1746-2851, HESS J1745–290) detected near the  _{in a com-}

prehensive manner.

We report here a complete study of the 2003-2004 Inte- gral/Ibis data of the  aiming to clearly depict the morphol- ogy of this interesting region at energies 20 to 400 keV, and to present the properties of the central source IGR J17456–2901.

The Integral observatory monitored the  region for all of 2003 and 2004, including some dedicated programs specifically planned to study the properties of the  . An important goal of this study was to search for correlated variability between the X-rays from Sgr A

and the higher energy emission from the central Integral source.

We describe the observations and data reduction methods in sections 2 and 3. Our results are presented in § 4 and include those of the multi-wavelength campaign on Sgr A

performed in 2004 pertaining to Integral ( § 4.4). Finally, we discuss some of their implications in § 5.

2. OBSERVATIONS

The Integral observatory carries two main γ-ray instruments;

Ibis and Spi, working in the energy range 15 keV to 10 MeV (Winkler 2003). Since its launch, Integral has observed the central degrees of our Galaxy for a total of about 7 Ms. This

TABLE 1 OL

Obs Start time Obs End time Pointings Exposure

(UT) (UT) (Ms)

2003 Spring^a 2003-02-28 2003-04-22 413 0.67

2003 Fall^b 2003-08-18 2003-10-16 805 2.20

2004 Spring 2004-02-07 2004-04-21 550 1.01

2004 Fall 2004-08-28 2004-09-17 262 0.53

aLoosely used to designate the first part of the year

bUsed to designate the second part of the year

time was divided between the Galactic Center Deep Exposure and Galactic Plane Scan core programs, and Guest Observer (  ) observations. In particular we include here analysis of

 programs performed in 2003 and in 2004 specifically ded- icated to the  _{. The 2004}  _{program ( ∼} 600 ks) was part of a broad multiwavelength campaign driven by a XMM-Newton large project aimed to study the flaring activity of Sgr A

.

The data that form the basis of this paper constitute a sub- set of all these observations selected such that the aim point is within 10

of the central  . We have performed a de- tailed analysis of 2174 pointings carried out between the end of February 2003 and October 2004. Each pointing or sci- ence window (ScW) typically lasts between 1800–3600 s dur- ing which the telescopes are aimed at a fixed direction in the sky. Table 1 gives a summary of the overall periods spanned by the observations.

With a total effective exposure time of 4.7 Ms at the position of Sgr A

, we have constructed high signal-to- noise images of the central degrees and the spectrum of the central source IGR J17456–2901 first detected by Inte- gral in 2003 (B´elanger et al. 2004). The results presented here are based on data collected with the Ibis/Isgri telescope (Ubertini et al. 2003; Lebrun et al. 2003) sensitive in the en- ergy range between 15 and 1000 keV. The angular resolution of the high-energy (15 keV–10 MeV) Spi telescope ( ∼ 3

) is not sufficient to resolve the contribution of the high-energy sources known to be present in the central degrees of the Galaxy and we have therefore not used these data.

The X-ray monitor on-board Integral, Jem-X, has a smaller field of view than Ibis and Spi and therefore the effective ex- posure at the location of Sgr A

is substantially lower ( ∼ 350 ks for the dataset considered in this paper). We analyzed the Jem- X data to produce mosaics for the  and discuss the results below.

We have also made use of XMM-Newton data from the multi- wavelength campaign on Sgr A

carried out in 2004 to con- struct the broadband high-energy spectrum of IGR J17456–

2901, and to identify the X-ray and soft γ-ray components. A detailed description of the XMM-Newton observations during this campaign, and of the characteristics of the two factor-40 flares detected from the direction of Sgr A

are presented in B´elanger et al. (2005).

3. DATA ANALYSIS METHODS

The basic Ibis/Isgri data reduction for individual point- ings was performed using the Integral Off-line Scientific Analysis software (  ) version 4.2 delivered by the 

(Courvoisier et al. 2003) and whose algorithms are described in Goldwurm et al. (2003b) and Gros et al. (2003). Following the recommendations related to the use of the  _software,

we restricted our analysis to events with energies greater than

20 keV. A number of additional procedures were implemented

in order to maximize the quality of our analysis given the large data set and the complexity of Galactic nuclear region. For in- stance, the analysis was consistently done twice: a first time to make a catalog of detected sources and perform a preliminary evaluation of the quality of each sky image, and a second time to ensure that all known sources in a given field of view were modeled correctly in the reconstruction of each sky image. The total number of sources detected by Isgri within 10

of the  _,

and therefore included in the analysis input catalog is 80. The background maps were constructed from empty field observa- tions at high latitudes in 256 bands from 17 to 1000 keV and in- corporated in the standard analysis where they were combined to match the chosen energy ranges. This procedure has allowed us to achieve the best possible quality for individual sky images and thus for the mosaic and spectrum. Other additional analysis procedures are detailed in the sections that follow.

3.1. Sky maps, light curves and mosaics

For a coded mask instrument like Ibis, a sky image is ob- tained by convolving the detector image or shadowgram with a decoding array derived from the spatial characteristics of the mask. For each sky image there is an intensity map (I), a vari- ance map (V) proportional to the total counts recorded in each pixel, and a significance map (S ) constructed from these as S = I/ √

V. Given that source photons are a very small frac- tion of the total counts, the background heavily dominates and the histogram of significance values for a given deconvolved image should follow the standard Normal distribution. Sig- nificant detections appear as spikes in the positive tail of the distribution. A broadening of this distribution is caused by sys- tematic effects unaccounted for in the standard analysis soft- ware. Therefore, the standard deviation of the distribution of significance values can be effectively used to characterize and quantify the quality of a given sky image, and must be taken into account when calculating the true detection significance of a signal in that image

. For this reason, we have weighted each sky image according to the variance of its significance distribu- tion to produce sky maps in seven energy bands from 20 to 400 keV; namely 20–30, 30–40, 40–56, 56–85, 85–120, 120–200, 200–400 keV, and to construct the spectrum of the  _source

IGR J17456–2901. We also performed the analysis of another energy band from 500 to 522 keV which yielded upper lim- its compatible with the Spi detection of the electron-positron 511 keV annihilation line reported in Knodlseder et al. (2005) (see § 4). The mosaics were made using the pixel spread option in the  4.2 imaging procedure which by projecting and dis- tributing pixel values on the final mosaic pixel grid, preserves the symmetry of the point spread function (  ) and allows the most accurate source positioning.

Since the central source is too weak to be fitted with the 

in the imaging procedure performed on each pointing, the light curve was obtained by extracting the count rate at the source’s pixel position from each reconstructed image.

3.2. Source position determination

The densely populated region around Sgr A

contains at least eight sources detected by Isgri within 1 degree of the  _{. Since}

† The method used to account for systematic effects on a per-ScW basis described here is equivalent to deriving a correction factor to the theoretical statistical uncertainty and applying it to each pixel in the variance map. It follows that any test statistic such as Pearson’sχ²-test in which the variance of each data point is used to weigh the associated data value when testing for a deviation from the mean, should be used with the corrected variance values, i.e. the statistical variance multiplied by the variance of the distribution of significance values.

the standard  pipeline does not perform simultaneous fitting of several sources, we have used a custom fitting procedure to determine the best fit positions of the sources in the neighbor- hood of the  . Sub-images 40 × 40 pixels in size centered on the radio position of Sgr A

were extracted from the mosaics in the different energy bands and fitted using a model that included up to eight point-sources, each characterised by a 2D Gaussian approximation of the system point spread function and applied in the  software (Gros et al. 2003)

The width of the  was left as free global parameter (1 for all sources) because for mosaics, the width of the final 

cannot be predicted. The presence of close sources and possi- ble confusion does not influence the result since a very strong point-like source (1E 1740.7–2942) dominates the images and this parameter is basically determined by the fit of this source.

The procedure gives us the possibility to fix the position of some sources and also to set a flat background level to be fitted together with the point-sources. The residual map is inspected to verify that the fitting procedure is performed correctly.

Statistical errors at 90% confidence level on the fitted posi- tions were derived from the measured source signal-to-noise ratios using the empirical law determined through a system- atic study of well known sources (Gros et al. 2003). Although the empirical function of the source location accuracy in terms of the detection significance was determined using images re- constructed from single pointings, it was recently validated on mosaics as well. These tests were performed on isolated point- sources and since we are studying a region where the source density is unusually high, we present the results of our simulta- neous fit for all the sources in the field to show that the offsets from the known positions are compatible with the empirical point-source location accuracy (  _).

3.3. Spectral extraction

The standard Isgri spectral extraction software works on the basis of a single pointing in the following manner. Using a list of sources, the procedure first calculates the pixel illumination factor; a model of the shadow of the mask cast on the detector plane by a given source. With this set of modeled shadow- grams, the software then attempts to fit the detector image by adjusting the relative intensity of each source either together with the background level simultaneously or by subtracting a background map normalized to the mean count rate before- hand. These procedures therefore estimate the maximum like- lihood intensity of each source and in each pointing. However, a difficulty arises in the case of week sources ( ∼ mCrab) that require very long exposure times to acquire sufficient statis- tics to be detected. In such cases, and in particular when a large number of sources have to be modelled, the most effi- cient way to derive a spectrum is to build and clean sky im- ages in the desired energy bands and then to extract in each the intensity and its associated variance at the pixel that corre- sponds to the source position, and to calculate the weighted mean count rate. As was described in the previous section, the variance at a given pixel must be corrected to take into ac- count the systematic effects that grow additively with exposure time. In the case of IGR J17456–2901, the flux and variance values were taken at the pixel that corresponds to the sky coor- dinates R.A. = 266.4168, decl. = –29.0078 (J2000) in each sky map; the position of Sgr A

.

3.4. X-ray spectrum

The X-ray spectrum in the range 1–10 keV was extracted us-

ing 2004 XMM-Newton data from a circular region centered on

the position of IGR J17456–2901 and with a radius of 8 arcmin-

utes. This radius was chosen to match the 13

 _{full width}

half max (  ) of the Ibis/Isgri instrument derived from the quadratric sum of the projected pixel (5

) and mask element (12

) sizes (Gros et al. 2003).

For a spectral extraction in which a large portion of the cam- era’s field of view is used as a collecting surface, it is most suitable to use XMM-Newton background event files compiled from high latitude observations to construct a background spec- trum. The procedure can be summarized as follows. Epic  _,

Mos 1 and Mos 2 event files are filtered to exclude all non X- ray triggers using the event flag and pattern, after which a good time interval (  ) selection is performed to exclude periods of solar flaring activity. The  selection criteria are based on the count rates in the high-energy bands, i.e.: 10–12 keV for Mos 1 and Mos 2, and 12–14 keV for  . An interval of 100 s qualifies as a  if it has less than 18 cts for both Mos cameras and 22 cts in the case of the  camera. These strict selection criteria en- sure that only the cleanest parts of the observation are used.

The resulting filtered event files are used to make images in the high-energy bands in which a uniform distribution of events is expected under the assumption that energetic charged particles heavily dominate the instruments’ respective spectra at these energies and that these fall uniformly on the telescope. If no point-sources are visible in these images, the ratios of the aver- age count rate in the images to that of the background event file are used to scale the background spectra.

4. RESULTS

We now come to the results we have obtained on the mor- phology of the Galactic nuclear region and on IGR J17456–

2901, which we tentatively associated with the supermassive black hole Sgr A

in B´elanger et al. (2004), and whose fea- tures we investigate more thoroughly in the present paper. We use three means of investigation to study the various charac- teristics of the source. The mosaic provides the fine position- ing and general shape of the emission from the source and its close neighbours. The individual sky maps provide the ele- ments needed for a variability study from kilosecond to month time scales and the average spectrum of the source can be used to constrain the nature of the emission. Section 4.1 begins with a presentation of the results obtained from the mosaic in the range 20–40 keV on the morphology of the emission from the central degrees. This is followed by a discussion of the changes in the emission’s morphology as a function of energy by look- ing at the mosaics in the different energy bands up to 85 keV, and ends with our results on the electron-positron annihilation line at 511 keV. In § 4.2 we discuss the light curves and vari- ability of the central source on different timescales, and in § 4.3 present the broad-band high-energy spectrum of the central ar- cmins of the Galaxy. Preliminary results on the  _{with the}

X-ray monitor Jem-X are briefly discussed.

4.1. Mosaics and spatial characteristics

The mosaic shown in Figure 1 was constructed by summing 2174 sky images from individual pointings and amounts to an effective exposure time of 4.7 × 10

s at the position of Sgr A

. This Figure presents the highest signal-to-noise Ibis/Isgri 20–

40 keV image of the  yet published, showing an excess of more than 45 in siginificance from the direction of Sgr A

.

To model the observed morphology we have assumed that the emission is due to the sum of the known high-energy point-sources of the region that have been detected by Inte- gral at least once. The main sources are 1E 1740.7–2942, KS 1741–293, 1A 1742–294, SLX 1744–299/300, 1E 1743.1–

2853, IGR J17456–2901 and SAX J1747.0–2853. The respec-

tive positions of these sources were derived from a simultane- ous fit of all 8 sources in the 20–40 keV mosaic. All positions were left as free parameters except for that of SAX J1747.0–

2853 that was fixed. This source was quite active for the period refered to as Spring 2004 and thus contibutes to the emission near the  but since its global contribution is weak and it can- not be clearly resolved from 1E 1743.1–2853 we must fix its position.

The result of the fitting procedure is well illustrated in Fig- ure 2 where we see the mosaic (left), the model (middle), and the residual map after subtraction of the model from the sky map (right). We can see that the spatial distribution of the mod- eled image resembles very closely that of the mosaic, even if the residues hint at the presence of a non-uniform underlying emission that is not properly taken into account in the tested model. The fitted source positions are listed in Table 2 where we also report the signal-to-noise, the estimated error radius corresonding to the 90% confidence level and the offset with respect to the proposed counterpart.

We find that the position of IGR J17456–2901, detected at a signal-to-noise level of 45 in this energy band, is R.A. = 17

45

42

.5, decl. = − 28

59

28

(J2000) with an un- certainty of 0

.75. The reliability of the derived position for this excess is supported by the fact that all the other sources in the field are very well positioned. The bright- est source, 1E 1740.7–2942, is well within its associated er- ror radius. For 1A 1742–294, the reported offset is very close to the value of  , while in the case of KS 1741–

293 the best known coordinate position itself has an uncer- tainty of about 1

(Sidoli & Mereghetti 1999). The source we labeled SLX 1744–299, is in fact a system composed of two known X-ray bursters located within ∼ 3

of each other (Skinner et al. 1990; Pavlinsky et al. 1994; Sakano et al. 2002) and for this reason we do not expect the fit to yield a position within the  for either one of the two sources, independently of the detection significance. 1E 1743.1–2853, a well known, bright X-ray source (Porquet et al. 2003a) is almost certainly contributing to the high-energy emission in the region. How- ever, performing the simultaneous fit using a single source to model the emission from the region around this source gives a centroid offset by about 2

from the XMM-Newton position known to arcsecond accuracy; a result that is not compati- ble with the expected error. Since, as mentioned above, we know of one source detected by Isgri that was active over the course of the first part of 2004, namely SAX J1747.0–2853, we included it and fixed its position. This yields a fitted po- sition for 1E 1743.1–2853 that is well within the uncertainty derived from the source’s detection significance. The position of IGR J17475–2822 was compared to the center of the Sgr B2 complex and discussed below.

IGR J17456–2901 is located at 1

.1 from the radio position

of Sgr A

and 0

.9 from the center of Sgr A East. It is there-

fore compatible with either of these sources. Indeed, even

if its associated positional uncertainty of 0

.75 is somewhat

smaller than the offset, we expect this  to be slightly over-

estimated when fitting multiple close sources (i.e. within the

full width of the  ). For example, in the case of the known

source 1A 1742–294, the measured offset can be 20–30% times

larger than the  . Morever we have found that the posi-

tions of 1E 1743.1–2853 and IGR J17456–2901 can change by

0

.3–0

.4 depending on the model adopted, and in some cases

IGR J17456–2901 is positioned only 0

.6 from Sgr A

. For this

reason we adopt a final error radius for the central source (and

for 1E 1743.1–2853 that has a comparable signal-to-noise ra-

tio) of 1

, about 30% larger than the  _{value of 0}

.75 derived

Sgr A*

SLX 1744-299 SAX J1747.0-2853

KS 1741-293 IGR J17475-2822 1E 1743.1-2843

1E 1740.7-2942

1A 1742-294 G0.9+0.1

F. 1.— Ibis/Isgri significance mosaic in the 20–40 keV energy range constructed from 2174 individual pointings with an effective exposure time at the position of Sgr A^∗of 4.7 Ms. Black indicates a statistical significance below or equal to 3σand white to a significance greater or equal to 60σ. Contours mark iso-significance levels from 9.5 to 75 linearly. The orientation is in Galactic coordinates. The grid lines indicate galactic coordinates with a spacing of 0.5 degrees.

F. 2.— Mosaic (left), model constructed from simultaneous fitting procedure of the 8 point-sources labeled in Figure 1 (center), and residuals after subtraction of model from mosaic (right). These maps are oriented in Equatorial coordinates where North is toward the top and East toward the left.

from the relation given by Gros et al. (2003).

On the other hand, we can safely exclude a number of other candidates such as the transient Asca source AX J1746.5–2901 (Sakano et al. 2002) mentioned by Revnivtsev et al. (2004) as a possible counterpart for the central excess based on the fact that it is located at a distance of more than 2

from it.

A similar analysis was performed on the mosaics from the same data set in different energy bands. As is clearly seen from the iso-significance contours in Figure 3, the morphology of the central degrees does not radically change with increasing en- ergy. However, we notice that the emission that seems to bridge the sources labelled Sgr A

and 1E 1743.1–2843 at low-energy persists at higher energies such that in the 56 to 85 keV range, the emission from the region seems to be centered between the two sources. This is a surprising result that we cannot readily interpret. An investigation of this based on a comparison of the emission detected by Ibis/Isgri with the 20 cm radio map, the 6.4 keV Fe line-emission contours and the CS map of the region

raises several other interesting questions.

Figure 4 is a radio continuum map at 20 cm (Yusef-Zadeh et al. 2004) on which we have overlayed the 20–30 keV iso-significance Isgri contours as they are shown in Figure 3 (top left). Firstly, the centroid of the very bright Sgr A complex which includes the luminous Sgr A East appears to be in best agreement with the 20–30 keV Isgri contours (Fig. 3 top left). For completeness, we performed the simultaneous fit using the position of Sgr A East instead of the one for Sgr A

and obtained very similar results. The offset of the fitted source is nonetheless slightly smaller for Sgr A East that for Sgr A

but both are within the  _and

thus statistically equivalent. We also see that the radio Arc

is quite distant from the peak near Sgr A

and can therefore

be confidently excluded as a possible contributor to the flux

at that position. However, the rough alignment of the radio

Arc with the elongation on the 20–30 keV contours in the

direction of negative latitudes is intriguing. In the 56–85 keV

TABLE 2

P Integral  G  

Source ID Signif. Fitted position



(R.A., decl.) (arcmin) (arcmin) 1E 1740.7–2942 . . 241.8 265.9794,−29.7430 0.28 0.14 1A 1742–294 . . . . 98.6 266.5138,−29.5109 0.40 0.55 SLX 1744–299 . . . 61.8 266.8600,−30.0183 0.60 1.14 KS 1741–293 . . . . 63.9 266.2130,−29.3327 0.59 1.23 1E 1743.1–2843 . . 46.3 266.5782,−28.7378 0.74 0.54 Sgr A^∗. . . 45.4 266.4285,−28.9918 0.75 1.13 IGR J17475–2822 . 18.9 266.8422,−28.4139 1.24 1.51 SAX J1747.0–2853 16.5 266.7500,−28.8700 1.45 0.0

aPoint-source location accuracy at 90% confidence level

bDistance between the fitted and nominal source positions

cPosition was fixed in the fit

contours (Fig. 3 bottom right) we see that the centroid of the emission is almost exactly between the Sgr A complex and the radio Arc. This region is known to harbour large molecular clouds and there appears to be a very good agreement between this high-energy emission feature and both the 6.4 keV Fe line-emission, tracing irradiated molecular regions, and the CS map, tracing regions with high gas density. Furthermore, the centroid of this high-energy source is strikingly close to that of the unidentified Egret source 3EG J1746–2851 and could in fact be its soft γ-ray counterpart. We extracted a spectrum for this source by fitting three sources, two of which had their positions fixed to those of Sgr A

and 1E 1743.1–2853, and obtained a power-law photon index of Γ ∼ 2.3, and a luminosity of L

∼ 2.6 × 10

erg s

for a distance of 8 kpc.

A more detailed analysis of this source will be presented in future work.

Moving in the direction of positive longitudes, we clearly see an emission region depicted in the 20–30 keV contours and whose centroid is labeled as IGR J17475–2822; a source associated with Sgr B2 by Revnivtsev et al. (2004). In- deed this source coincides with the radio-bright Sgr B2 com- plex composed of molecular clouds and several compact HII regions. The fitted centroid for this source, taken to be point-like in the 20–40 keV image, is positioned 1

.6 from the estimated center of the cloud. As was pointed out by Revnivtsev et al. (2004), there is good agreement between the 6.4 keV Fe line contours and the emission detected by Is- gri as IGR J17475–2822. Moreover, the extension toward the north could tentatively be associated with the composite 

G 0.9+0.1 (Helfand & Becker 1987). The X-ray emission of its pulsar wind nebula (  ) has been mapped with XMM- Newton (Porquet et al. 2003) and the extrapolation of the flux towards 20 keV of about 9 × 10

ph cm

s

, is consistent with the residual flux of roughly 0.06 cts/s (20–40 keV) or 8 × 10

ph cm

s

. It is therefore tempting to interpret it as a detection of the highest energy synchrotron radiation from this source that was detected for the first time this year by the Hess instrument (Aharonian et al. 2005a). It is worth noting that a long XMM-Newton exposure of this object has revealed the presence of variable source probably of an accreting binary type, located at a distance of 1

(Sidoli et al. 2004). Having a luminosity close to that of G 0.9+0.1, its contribution to the residual Isgri emission could be significant.

We performed the analysis of the entire data subset used to construct the mosaics in the narrow band between 500 and 522 keV. This corresponds to the  of the emission line in the Isgri instrument (Terrier et al. 2003). A background map

that corresponds to this energy range was used and thus the resulting mosaic is free of systematic effects. No sources are detected in the field spanning 10 degrees on either side of the

 both in longitude and latitude. We obtain a 3σ upper limit of 1.9 × 10

ph cm

s

to the flux from a point-source at the position of Sgr A

where the exposure and thus the sensitiv- ity is maximal. This limit is calculated taking into account corrections derived from the probability of photoelectric in- teraction at 511 keV in the Isgri detector (34%) and the fact that we have selected events using a band corresponding to 78% of line flux. Spi detected a 511 keV line flux of about 10

ph cm

s

with intrinsic line width of 2.7 keV (  ₎

from a region well described by a Gaussian with a  _of

about 8

and that coincides approximately with the Galactic bulge (Kn ¨odlseder et al. 2005). If we assume that our sensitiv- ity is more or less uniform over the central 10 degrees around Sgr A

, our upper limit implies that if this emission is due to a collection of n point-sources clustered together such that they cannot be resolved by Spi, then under the simplifying assump- tion that they all contribute equally to the total flux, each must have a flux of about

× 10

ph cm

s

. Therefore, at least 5 individual sources would be necessary to account for this ex- tended 511 keV emission in general agreement with the Spi result (Kn ¨odlseder et al. 2005).

Finally, Jem-X mosaics in four energy bands, with a total ex- posure is 3.2 Ms and effective exposure at the location of Sgr A

of about 500 ks, were constructed from 1204 science windows taken between 2003 February to 2004 October. We used the following energy ranges: 3–4, 4–8, 8–14, 14–35 keV and find no evidence in this data sample for the presence of a Jem-X source in the Sagittarius A complex except for a very marginal excess in the 8–14 keV mosaic. Although this analysis is pre- liminary and at this point somewhat qualitative, it is an interest- ing result in the light of the strong Isgri detection and obvious intense X-ray emission from this region seen by XMM-Newton and Chandra. It may be an additional indication that the emis- sion is not due to a point-source but rather to a compact diffuse emission region were thermal and non-thermal processes take place.

4.2. Light curves and variability study

The complete light curve of IGR J17456–2901 in the 20–

40 keV energy range, with a resolution of about 1800 s cor- responding to the duration of a single pointing, is shown in Figure 5. Since low amplitude variability on kilosecond time scales can not be meaningfully studied for such a weak source due to statistical limitations, we have also done a search on longer time scales by rebinning the total light curve on the ba- sis of 1 day, 2 weeks and 1 month. These rebinned data sets are presented in Figure 6.

No individual point deviates from the mean by more than 3σ. The level of variability in the flux from the central source was evaluated by means of a simple chi-squared test. For the unbinned data set shown in Figure 5, the reduced chi-squared value is χ

= 1.3 (2758/2093). For the light curve with 1-day time resolution (Fig. 6 top) we found χ

= 1.7 (180/109). In the case of the 2-week time resolution light curve (Fig. 6 middle), the reduced χ

value was found to be 3.6 (61/17). However, if we exclude the first point in this data set which corresponds to the data collected during revolution 46, the first observation of the  just after the initial calibration phase, we find a value of χ

= 2.1 (34/16), in closer agreement with the previous two.

Finally, in the case of the 1-month time resolution light curve

(Fig. 6 bottom) we find values of 5.2 (52/10) and 3.1 (28/9) if

we exclude the first data point, heavily affected by the revolu-

Sgr A*

SAX J1747.0-2853

KS 1741-293 IGR J17475-28221E 1743.1-2843

1E 1740.7-2942

1A 1742-294 G0.9+0.1

Sgr A*

SAX J1747.0-2853

KS 1741-293 IGR J17475-28221E 1743.1-2843

1E 1740.7-2942

1A 1742-294 G0.9+0.1

Sgr A*

SAX J1747.0-2853

KS 1741-293 IGR J17475-28221E 1743.1-2843

1E 1740.7-2942

1A 1742-294 G0.9+0.1

Sgr A*

SAX J1747.0-2853

KS 1741-293 IGR J17475-28221E 1743.1-2843

1E 1740.7-2942

1A 1742-294 G0.9+0.1

F. 3.— Ibis/Isgri significance mosaic as in Figure 1 in four energy bands: 20–30 keV (top left), 30–40 keV (top right), 40–56 keV (bottom left) and 56–85 keV (bottom right). Black corresponds to statistical significance below or equal to 3 and white to a significance greater or equal to 50. The 12 contours levels mark iso-significance from 8 to 70 linearly.

Sgr A*

SAX J1747.0-2853

KS 1741-293 IGR J17475-2822 1E 1743.1-2843

1E 1740.7-2942

1A 1742-294 G0.9+0.1

F. 4.— Radio map of the Galactic center region at 20 cm overlayed with the 20–30 keV Isgri contours.

tion 46 data given that there is a three-week gap between this revolution and the second observation of the  during revolu- tion 53. These reduced chi-squared values tend to increase as the binning gets coarser and thus we might be seeing a small level of variability on monthly timescales. Disregarding the data point associated with rev. 46, the only deviation that al- most reaches 3 σ from the mean is at the very end of the light curve where. This lack of evidence for significant variations in flux other than the low level of variability seen on monthly timescales is in contradiction with the previous detection of a flare from IGR J17456–2901 (B´elanger et al. 2004) that we therefore do not confirm. We point out that those results were obtained with the preliminary analysis procedures and without background corrections. The data subset covering the obser- vation period of the reported flare (2003 April) processed with the most recent analysis software and background correction maps do not indicate significant variability with respect to the mean count rate. Similarly, the sources 1E 1743.1–2843 and IGR J17475–2822 seem rather constant unlike the four well

known X-ray binaries that show very large intensity variations over the two-year observation period.

4.3. Spectrum of IGR J17456–2901

Figure 7 shows the Isgri spectrum of the  source that we modelled with a simple power-law of index Γ = 3.04 ± 0.08 (χ

= 7.92 for 5 dof and 3% systematics). The pegged power-law model pegpwrlw in Xspec uses the total flux as nor- malization and in this way the photon index and normalization are independent paraters. The total flux in the range from 20 to 400 keV is F

= (7.02 ± 0.27) × 10

erg cm

s

, which corresponds to a luminosity of L

400 keV]

=(5.37 ± 0.21) × 10

erg s

at a distance of 8 kpc to the  . In the 20–100 keV range, the luminosity is L

100 keV]

= (4.56 ± 0.10) × 10

erg s

, somewhat higher than our first estimate of ∼ 3 × 10

erg s

(B´elanger et al. 2004).

This is not surprising given that the first estimate was based

on a rough comparison with the Crab’s count rate in only two

energy bands, 20–40 and 40–100 keV, and that we now have

F. 5.— Light curve of IGR J17456–2901 in the 20–40 keV band constructed from 2174 sky images, each corresponding to one pointing (1800 s).

F. 6.— Rebinned light curves of IGR J17456–2901 constructed from the data set shown in Figure 5 with time bins of 1 day (top), 2 weeks (middle) and 1 month (bottom).

5 points to constrain the slope. Furthermore, the detection significance of the central source was much lower than in the present case.

Now turning to the broad-band high-energy spectrum of IGR J17456–2901, we can see in Figure 8, the spectrum of the central source from 1 to 400 keV where the X-ray portion (1–

10 keV) is from XMM-Newton data collected during the mul- tiwavalength campaign, and therefore contemporaneous with part of the Isgri data from 2004 used to construct the soft γ- ray portion (20–400 keV) of the spectrum that was discussed earlier and is shown by itself in Figure 7. The X-ray spectrum was made by extracting the photon flux from a region centered at the position of IGR J17456–2901 and with an radius of 8

. This integration radius was chosen to be compatible with the Ibis/Isgri psf because there is no obvious X-ray point-source counterpart to IGR J17456–2901 within 1

of Sgr A

. Such a point-source would have to be hard, persistant and extremely bright in X-ray in order to be compatible with the high-energy flux of the Integral  _source.

The model fitting for large extended regions near the  _is

challenging for two main reasons. First, the X-ray spectra of such extended regions give a corse, averaged spectral be-

F. 7.— Two-year averaged Isgri spectrum of the



2901 in the range 20–400 kev. The spectrum is fit with a power-law normalized over the whole energy range and yields a photon index ofΓ =3.04±0.08.

Total flux is (7.02±0.27)×10⁻¹¹erg cm⁻²s⁻¹. The last point in the spectrum corresponding to the range 200–400 keV is given as the 1σupper limit, and the data point in the 120–200 keV band has a significance∼2σ

haviour of a complex field heavily dominated by diffuse emis- son that we know to have several different spectral components (Muno et al. 2004a) but that also includes all the point sources some of which surely contribute to the hard X-ray flux. Of course, this difficulty dissipates as integration radius decreases since fewer components are summed together. Second, we have no a priori knowledge of the nature of the emission detected as IGR J17456–2901 and therefore do not know whether the comparison with the total X-ray spectrum from a region that roughly corresponds to Isgri’s angular resolution is an appro- priate one. Keeping these caveats in mind, we justify this type of comparison by pointing to the fact that the source coinci- dent with the  , detected by Integral as what appears to be a point-source, must undoubtedly contribute to the X-ray spec- trum from the region that corresponds to its spatial extent. The spectral transition from 10 to 20 keV must be more or less con- tinuous and therefore we expect the high-energy component present in the X-ray spectrum and from which IGR J17456–

2901 arises, to stand out beyond the thermally dominated spec- trum at around 20 keV. Therefore, a large χ

value should not be surprising for it points to the fact that the emission in the range from 1 to 3 keV is not modelled properly for the rea- son mentioned above. Our aims in this section is to constrain the high-energy characteristics of this source in the range 1–

400 keV using the Isgri spectrum above 20 keV.

The broad-band spectrum can be modelled using a simple broken power-law over the entire range to get an idea of the change of spectral index with increasing energy. However, in order to be as constraining as possible without overlooking pos- sibly important components to this emssion like the hot temper- ature plasma present in the  region, we performed the fit with the same model as the one used by Muno et al. (2004a) to fit the diffuse emission from the various regions in the 17

× 17

field around Sgr A

referred to as Southeast, Southwest, North- west, East, Close, and Northeast by the authors. This model comprises a two-temperature plasma with different absorption columns, a power-law and a gaussian line to fit the 6.4 keV neu- tral Fe emission line absorbed with the same column density.

Although providing a reasonable fit to the data from 2–8 keV, the model does not work well in the Isgri range of the spectrum.

The power-law fit with index Γ ≈ 2 underestimates the flux in

the 20–40 keV range and overestimates it above 85 keV. A

somewhat better fit is provided by replacing the simple power-

law with either a cutoff or broken power-law. In the case of the

TABLE 3

SM  GCS

IGR J17456–2901

Quantity CutoffPL Broken PL

NH,1(10²²cm⁻²) . . . 7.81^+0.02

−0.04 7.79^+0.11

−0.13

kT1( keV) . . . 1.002^+0.008

−0.004 0.99^+0.02

−0.02

NkT₂ . . . 0.378^+0.009

−0.006 0.38^+0.02

−0.02

NH,2(10²²cm⁻²) . . . 13.13^+0.17_−0.23 13.52^+0.56_−0.50 kT2( keV) . . . 6.56^+0.07_−0.09 6.60^0.13_−0.12 NkT₂(10⁻²) . . . 6.35^+0.05_−0.06 6.45^+0.01_−0.02 Γ₁ . . . 1.09^+0.03

−0.05 1.51^+0.06_−0.09 Ecutoff/break( keV) . . . 24.38^+0.55

−0.76 27.13^2.79_−4.39 Γ₂ . . . · · · 3.22^+0.34

−0.30

NΓ(10⁻³ph cm⁻²s⁻¹keV⁻¹) 4.46^+0.29

−0.27 7.445^+0.001

−0.001

χ²(dof) . . . 4490.7 (2658) 4458.0 (2657) Uncertainties on the parameters correspond to the 90% confidence level

cutoff power-law, we found a photon index Γ ≈ 1 and cutoff en- ergy of about 25 keV, and in the case of the broken power-law, the photon indeces were found to be Γ

≈ 1.5 and Γ

≈ 3.2 with a break energy of about 27 keV. The best fit parameters values are given in Table 3 where only the free parameters are listed;

all abundances are fixed to solar abundance.

Looking closely at the unfolded spectra shown in Figure 8, we can distinguish the low and high temperature plasma com- ponents drawn in red and green respectively, the gaussian line in light blue and the broken power-law in dark blue. The hot thermal component clearly dominates the spectrum at low- energies but its contribution is already well below that of the power-law component in the 20–30 keV band and is totally neg- ligible beyond that. If we fix the temperature of the hot com- ponent at 8 keV, the effect on the other parameters is small.

The photon index in the cutoff power-law decreases from 1.09 to 1.05 and the high-energy cutoff from 24.4 to 27.7 keV. The χ

value increases to 4849.0 for one more dof and therefore the reduced chi-squared is slightly larger i.e. χ

= 1.82, and the contribution of this component to the overall flux in the 20–30 and 30–40 keV bands increases by a factor of 2 but still lies 3 times below the power-law in the first band and a factor of 7 below in the second.

4.4. Multiwavelength campaign

A multi-wavelength campaign to study Sgr A

with a total exposure time of about 500 ks was performed in two segments, the first of which was from 2004 March 28 to April 1 and that we refer to as epoch 1, and the second from 2004 August 31 to September 3 refered to as epoch 2. The primary aim of this campaign was to study correlated variability, particularly in the IR, X-ray and soft γ-ray energy bands. Figures 9 and 10 show the 2–10 keV XMM-Newton light curve of a 10

region around Sgr A

in the bottom panels, and the 20–30 keV Isgri light curve of IGR J17456–2901 in the upper panels for epochs 1 and 2 respectively. As is clearly visible, the periods during which the factor-40 flares from the direction of Sgr A

occured do not have simultaneous coverage in the Integral data. Unfortunately, both data gaps in the Isgri light curve correspond to the period between orbits. For this reason we are still unable to conclude whether or not we can expect to detect a correlated variability in the X-ray and soft γ-ray bands. There are two features worth mentioning that can be noticed in Figure 10 although they have marginal statistical significance. First, two points in the Isgri light curve, approximately in the middle of the upper panel,

stand out at about 2.5 σ above the mean. These are tempo- rally coincident with the two hiccups at the end of the first data subset in the X-ray light curve shown in the bottom panel. Sec- ond, the weighted mean X-ray count rate is somewhat higher in the first data subset (0.30 ± 0.001 cts/s) than in the second (0.27 ± 0.001 cts/s), a behaviour apparently seen also in the two corresponding segments of the Isgri light curve where the weighted mean count rate is 0.42 ± 0.03 cts/s in the first and 0.20 ± 0.07 cts/s in the second. This indicates that there may be a relationship between the flaring activity of Sgr A

and emis- sion at higher energies. Future simultaneous observations will undoubtedly help elucidate this point which remains uncertain.

5. SUMMARY AND DISCUSSION 5.1. Summary

We have studied the morphology of the high-energy emis- sion from the central few degrees of the Galaxy in the energy range from 20 to 400 keV based on a sample of Integral data collected from 2003 February to 2004 October with a total live- time of 7 × 10

s. We paid particular attention to the character- istics of the emission from the Galactic nuclear region where we detect a source with high significance in the 20–40 keV en- ergy range located at R.A. = 17

45

42

.5, decl. = − 28

59

28

(J2000) with an uncertainty of 1

and therefore compatible with the position of the central black hole Sgr A

. This detection confirms the results obtained by B´elanger et al. (2004) on the

 source IGR J17456–2901.

The source IGR J17456–2901 is persistent and shows no variability at the 3σ level in contradiction with what was sug- gested to in B´elanger et al. (2004). This result holds at kilosec- ond, daily, bi-weekly and monthly timescales.

The spectrum of the central source in the 20–400 keV range is well fit by a power-law of index Γ ≈ 3. We have combined this dataset with the X-ray spectrum of a circular region with radius 8

centered the Integral  source derived from partially contemporaneous XMM-Newton data collected during obser- vations of the  performed in 2004 in the range 1–10 keV.

From this we find that the broad-band high-energy spectrum can be fit equally well with a model that comprises a two- temperature plasma (kT

≈ 1.0, kT

≈ 6.5), a Gaussian line to account for the neutral Fe emission at 6.4 keV, and either a cutoff power-law with photon index Γ ≈ 1 and cutoff energy of about 25 keV or a broken power-law with photon indeces Γ

≈ 1.5 and Γ

≈ 3.2, and break energy ≈ 27 keV.

We also detect hard (Γ ∼ 2.2) emission from a region located between Sgr A

and the radio Arc that seems to coincides with the 6.4 keV emission from neutral to weakly ionised Fe and with the CS map of the region. As is the case with IGR 17475–

2822, we believe that this new detection of hard X-ray emission orginates in one or several large molecular clouds known to exist in that region.

The nature of the emission from the direction of the Sgr A complex detected as IGR J17456–2901 is unknown. In what follows we discuss a number of scenarios in an attempt to iden- tify the source of the emission detected by Integral and appar- ently centered on Sgr A

.

5.2. Hot plasma

The hot component at 6–8 keV of the two-temperature

plasma at the  is well known and its presence is viewed by

many as problematic in terms of it being confined given the es-

cape velocity of a Hydrogen plasma at that temperature or of

the heating mechanisms that would be required to supply en-

ergy to the plasma were it not confined. It is interesting to ask

what is this hot plasma’s extrapolated flux at energies between